Endocrine Abstracts

Introduction: Fine needle aspiration biopsy (FNAB) is the mainstay diagnostic procedure for evaluation of thyroid nodules, but it doesn’t permit to distinguish between benign and malignant follicular lesions (category IV in the Bethesda Cytopathology System). Some reports have suggested an association between increased serum levels of TSH and thyroid cancer. However, the specific relationship between TSH and malignancy has been scarcely studied in follicular thyroid nodules.

Objectives: To investigate the association between preoperative TSH levels and malignancy in patients with follicular thyroid nodules.

Methods: Retrospective study of all subjects who underwent surgical treatment for Bethesda IV thyroid nodules since the adoption of the Bethesda System for reporting thyroid cytopathology in a single centre from Spain (2012–2017), and for whom a preoperative measure of TSH was available. Cytopathologic diagnosis was established by ultrasound-guided FNAB. Clinical data and ultrasound characteristics of the nodules were recorded.

Results: A total of 125 subjects were identified (mean age 52.8±14.0 years, 87.2% women). Forty-nine of them (39.2%) showed malignancy upon definitive histopathological examination (21 follicular variant of papillary carcinoma, 19 follicular thyroid carcinoma, 8 other types of papillary carcinoma and 1 medullary carcinoma). Subjects with thyroid cancer were older than those with benign conditions (57.0±13.7 vs 50.1±13.7 years; P=0.007). The presence of ultrasonographic traits of malignancy (hypoechogenicity, irregular margins, microcalcifications, taller than wide shape and central vascularization), the maximum diameter of the nodule and its calculated volume didn’t differ among malignant and benign lesions. Median (range) levels of TSH were 2.35 mU/l (0.34–37.0) among subjects with thyroid cancer and 2.04 mU/l (0.09–10.94) among those with benign diseases (P=0.35). The ROC analysis showed a TSH cut off point of 2.14 mU/l to differentiate benign from malignant disease (sensitivity 65.3%, specificity 56.6%, positive predictive value 50.8%, negative predictive value 71.7% and area under the curve 0.55). The proportion of subjects with TSH ≥2.14 mU/l was greater among subjects with cancer (P=0.018). A logistic regression model including age and TSH ≥2.14 mU/l as independent variables showed that both of them retained independent association with malignancy (OR [95% CI] for TSH ≥2.14 mU/l: 2.53 [1.17–5.44]; P=0.018).

Conclusions: The present study supports an association between serum concentrations of TSH and risk of malignancy among subjects with Bethesda IV thyroid nodules. In this particular group of patients TSH levels could provide greater diagnostic yield than ultrasonographic examination.